6 results about "Robust least squares" patented technology

The application discloses a time difference positioning method based on random robust least square, mainly solves the threshold effect of the existing time difference positioning method when the measurement error increases, and leads to the problem of sharp deterioration of positioning accuracy, and the implementation scheme is as follows: a plurality of observation stations and noise measurement errors are set, the time difference positioning equation is obtained by using the position parameters and the measurement errors, an auxiliary variable is defined, the auxiliary variable is introduced into the time difference positioning equation to establish a constraint equation, the constraint equation is processed by using a random robust least square method to obtain a cost function, the cost function is solved by using a least square method to obtain a target position, a new constraint equation is established according to the internal relation of the auxiliary variable elements, the new constraint equation is solved by using a weighted least square method to obtain a positioning error, the target position is linearly corrected by using the positioning error to obtain a final positioning result, and the target positioning is completed. The application reduces the positioning error, improves the positioning accuracy and the robustness of positioning, and can be used in aerospace, navigation and electronic warfare.

The application discloses a kind of calculation method and system of detector parallelism, belong to optical detection technical field.The method includes: collecting at least four positioning points of lower surface of detector, at least three reference points of upper surface and upper surface full pixel point cloud data, and carry out de-duplication and coordinate calibration;Fitting lower surface plane;Based on the reference point of upper surface, rough fitting is carried out, and according to the distance of each pixel to rough plane, the prior weight reflecting physical credibility is generated;In the fine fitting of upper surface, the prior weight and the Huber dynamic weight based on residual adaptive calculation are iteratively fused, and the fine plane parameters are solved by weighted robust least square algorithm;Finally, the angle of upper and lower surface normal vector is calculated to obtain parallelism.The application cooperates to suppress outlier interference by double weight mechanism, and global trend and local accuracy are considered, which significantly improves the anti-interference and precision of parallelism calculation, especially suitable for precise installation and adjustment of high-resolution optical detector.

This invention relates to a novel method for analyzing fault characteristics in power systems. First, it collects and preprocesses three-phase voltage, three-phase current, zero-sequence electrical quantities, protection device operation information, and fault indicator information at the grid connection point and lines. Then, it performs multi-domain feature extraction to simultaneously acquire transient change features, steady-state sequence component features, frequency domain wavelet packet energy entropy features, and harmonic distortion features, thereby constructing a multi-dimensional fault feature vector. Next, it uses an improved attention mechanism fusion model to fuse the multi-dimensional fault feature vector, and based on the fused fault feature vector, constructs a deep neural network model. Finally, it employs a robust least squares algorithm to calculate fault impedance and locate the fault, outputting a complete analysis result of the fault characteristics. This result is then uploaded to the dispatch master station to generate fault waveform reports, characteristic curves, and maintenance suggestions. This invention has the advantages of multi-domain fusion, strong anti-interference capability, and accurate fault identification.

The invention relates to the technical field of power system operation and control, and discloses a voltage sensitivity-considered feeder adjustment characteristic modeling and parameter identification method, system and device and a medium, and the method comprises the following steps: fully considering the complex dynamic characteristics of feeder load through multi-step data processing and an advanced parameter identification method; in the aspect of data processing, environmental factor trend term power data are firstly removed, voltage power data strongly related to voltage fluctuation caused by load change are accurately extracted, and a high-quality data basis is provided for subsequent modeling. In the aspect of modeling, a step-down energy-saving coefficient model is established, and the adjustment potential of the feeder load can be accurately represented. And in a parameter identification link, a robust least square identification method based on random sampling consistency is combined, an optimal model is screened out through multi-round iteration, and model parameters are dynamically adjusted by adopting a sliding time window mechanism, so that the model can adapt to changes of load characteristics and operation conditions.

The application discloses a kind of unmanned aerial vehicle laser sounding point cloud filtering method based on scanning line and sliding window, it is related to point cloud data processing technical field, filtering method includes the following steps: first, read scanning line data and use radius filtering to separate outlier to bottom point cloud;Second, with scanning point time as X axis, elevation as Y axis, scanning line point cloud is converted into two-dimensional data and constructs sliding window;Use robust least squares to fit data in window, determine terrain complexity according to fitting median deviation, and then construct adaptive filtering method based on terrain characteristic value, complete single scanning line fine filtering by point-by-point sliding window;Finally, traverse all scanning lines and repeat the above steps.The application adopts the joint filtering method from "coarse to fine", principle is simple, convenient to operate, and applicable.The noise point is effectively separated while the terrain detail features are completely retained, which can provide support for unmanned aerial vehicle laser sounding in nearshore shallow water complex area mapping application.

This application discloses an automatic detection method and system for pulley runout in water pumps. The method includes: sequentially acquiring the angular position and displacement values ​​during the first forward rotation, reverse rotation, and second forward rotation of the pulley under the same clamping conditions; reversing the reverse displacement data and extracting low-order periodic components from both the reverse and first forward displacement data, and obtaining the backlash angle through cross-correlation to correct the angular coordinates of the reverse data; interpolating the three displacement data segments to the same angular grid, fitting the positional variation term based on the difference between the two forward data segments, and correcting the first forward displacement value accordingly; then using the corrected three displacement values ​​as input, solving for the common geometric components through robust least squares, and using the difference between the maximum and minimum values ​​within one revolution as the pulley runout value. This application can reduce the impact of commutation backlash, clamping positional variations, and abnormal sampling on the detection results, improving the repeatability and reliability of the detection.